T-cell responses are activated by short peptides derived from foreign proteins of pathogenic origin. Each pathogen consists of thousands of potential immunogenic antigens, but only a small fraction of these epitopes are recognized by CD8 T-cells during the course of an infection. This phenomenon, known as immunodominance, is fundamental property of T cell immunity (Yewdell and Bennink, 1999), and results from many factors that govern antigen presentation and T-cell activation (Yewdell, 2006). The large quantity of available immunological data, such as HLA binding measurements, has led to the development of powerful modeling tools that provide robust estimations of several processes that are part of the adaptive immune pathway, including MHC binding (Peters et al., 2006; Lin et al., 2008), proteasomal cleavage (Kesmir et al., 2002; Schatz et al., 2008), and Transporter Associated with antigen Processing (TAP) (Peters et al., 2003). Recent work has shown the viability of using such prediction tools in epitope mapping studies (Moutaftsi et al., 2006; Assarsson et al., 2008). However, developing computational tools to study immunodominance has not been explored, partially due to the lack of sufficient amounts of data to learn such models. We propose a computational approach for studying immunodominance both in terms of antigenic context and host genetics. We will develop and evaluate new computational tools for studying immunodominance. We will then use these tools to predict immunodominant responses, and to study its underlying mechanisms. The core of our approach involves the ability to predict HLA-peptide interactions, leveraging previously developed tools for this analysis. We demonstrate initial results that compare in vitro and in silico epitope mapping on a set of matched cases and controls from the Step HIV vaccine trial data. We also present initial results on the characterization of HLA-peptide interactions and their implications to the co-evolution of the immune system and pathogens it faces. We show how this tool can be used to discern highly immunogenic and highly non-immunogenic proteins encoded and expressed in an adenovirus vector used in a recent HIV vaccine trial. The disease focus of this proposal is HIV, and the tools will be developed and applied to HIV vaccine trial laboratory studies performed at the FHCRC HIV Vaccine Trial Network Laboratory Program. HIV remains one of the world's deadliest pandemics, with an estimated 2.5 million new infections last year and almost the same amount of AIDS deaths. Thus, with the great need for an HIV vaccine, the ongoing search for an effective vaccine is an intense focus of HIV research. By developing computational tools for predicting an individual's immune responses, we hope to gain insights into ways these responses can be altered, an advance which would help improve HIV vaccine design strategies as well as clinical trial design and analysis for HIV vaccines.